Lowering detector confidence to increase number of detections & boxes

Xavier agx dev kit
jetpack 4.3
deepstream4.02

How do I lower detector confidence threshold to increase the number of boxes and detections?

In the primary-gie config file have tried:

threshold=0.1
classifier-threshold=0.1

but it doesn’t seem to be any different to when they are =0.9

Is there somewhere else where the detection level can be set?

Thanks

There is a threshold for detectors pre-cluster-threshold

thanks bcao,

I only seem to be able to change the threshold for my detectnet model not my yolo models?

do I need to do something different with yolo models?

I’m using YoloV3Tiny


Update:

I think I need to change threshold parameters in nvdsinfer_custom_impl_Yolo/nvdsparsebbox_Yolo.cpp.

static bool NvDsInferParseYoloV3(
...
    const uint kNUM_BBOXES = 3;
    static const float kNMS_THRESH = 0.3f;
    static const float kPROB_THRESH = 0.7f;
...

I think these values are shared V3 and V3Tiny?

Can you tell me what kNMS_THRESH and kPROB_THRESH mean individually? what each one does?

I assume I need to create a new libnvdsinfer_custom_impl_Yolo.so file every time I want to adjust the sensitivity?

Thanks

There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

Do you mean the pre-cluster-threshold not work for yolo?

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.